{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7776935019376858,
  "global_step": 300000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.243,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.6562,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999999999999999e-06,
      "loss": 0.4946,
      "step": 3000
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.46,
      "step": 4000
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.6666666666666664e-05,
      "loss": 0.4355,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.4203,
      "step": 6000
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.3333333333333332e-05,
      "loss": 0.4075,
      "step": 7000
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.3986,
      "step": 8000
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 0.3922,
      "step": 9000
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.333333333333333e-05,
      "loss": 0.3856,
      "step": 10000
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.666666666666666e-05,
      "loss": 0.3801,
      "step": 11000
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.9999999999999996e-05,
      "loss": 0.3758,
      "step": 12000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.333333333333333e-05,
      "loss": 0.3712,
      "step": 13000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.6666666666666665e-05,
      "loss": 0.3671,
      "step": 14000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 0.364,
      "step": 15000
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.3605,
      "step": 16000
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.666666666666666e-05,
      "loss": 0.3574,
      "step": 17000
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.3553,
      "step": 18000
    },
    {
      "epoch": 0.11,
      "learning_rate": 6.333333333333333e-05,
      "loss": 0.3516,
      "step": 19000
    },
    {
      "epoch": 0.12,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.3494,
      "step": 20000
    },
    {
      "epoch": 0.12,
      "learning_rate": 7e-05,
      "loss": 0.346,
      "step": 21000
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.333333333333332e-05,
      "loss": 0.3432,
      "step": 22000
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.666666666666666e-05,
      "loss": 0.3415,
      "step": 23000
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.999999999999999e-05,
      "loss": 0.3385,
      "step": 24000
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.333333333333333e-05,
      "loss": 0.3367,
      "step": 25000
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.666666666666665e-05,
      "loss": 0.3344,
      "step": 26000
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.999999999999999e-05,
      "loss": 0.3322,
      "step": 27000
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.333333333333333e-05,
      "loss": 0.331,
      "step": 28000
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.3287,
      "step": 29000
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.3266,
      "step": 30000
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00010333333333333333,
      "loss": 0.3252,
      "step": 31000
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00010666666666666667,
      "loss": 0.3237,
      "step": 32000
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00010999999999999998,
      "loss": 0.3215,
      "step": 33000
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00011333333333333331,
      "loss": 0.3207,
      "step": 34000
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00011666666666666665,
      "loss": 0.3197,
      "step": 35000
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.3181,
      "step": 36000
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001233333333333333,
      "loss": 0.3178,
      "step": 37000
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00012666666666666666,
      "loss": 0.3163,
      "step": 38000
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00013,
      "loss": 0.3154,
      "step": 39000
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001333333333333333,
      "loss": 0.3146,
      "step": 40000
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00013666666666666666,
      "loss": 0.313,
      "step": 41000
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00014,
      "loss": 0.3131,
      "step": 42000
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00014333333333333334,
      "loss": 0.3115,
      "step": 43000
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00014666666666666664,
      "loss": 0.3111,
      "step": 44000
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00015,
      "loss": 0.3102,
      "step": 45000
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00014999833143445625,
      "loss": 0.3094,
      "step": 46000
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.000149993325817371,
      "loss": 0.3091,
      "step": 47000
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00014998498338737861,
      "loss": 0.3077,
      "step": 48000
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00014997330454219024,
      "loss": 0.3075,
      "step": 49000
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001499582898385749,
      "loss": 0.3067,
      "step": 50000
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00014993993999233321,
      "loss": 0.3054,
      "step": 51000
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00014991825587826286,
      "loss": 0.3048,
      "step": 52000
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00014989323853011714,
      "loss": 0.3043,
      "step": 53000
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00014986488914055563,
      "loss": 0.3036,
      "step": 54000
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00014983320906108733,
      "loss": 0.3025,
      "step": 55000
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001497981998020062,
      "loss": 0.3025,
      "step": 56000
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001497598630323192,
      "loss": 0.3022,
      "step": 57000
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00014971820057966672,
      "loss": 0.3016,
      "step": 58000
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00014967321443023538,
      "loss": 0.3012,
      "step": 59000
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014962490672866348,
      "loss": 0.3006,
      "step": 60000
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00014957327977793857,
      "loss": 0.3003,
      "step": 61000
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00014951833603928788,
      "loss": 0.3008,
      "step": 62000
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001494600781320608,
      "loss": 0.2995,
      "step": 63000
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00014939850883360407,
      "loss": 0.2988,
      "step": 64000
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001493336310791294,
      "loss": 0.2988,
      "step": 65000
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00014926544796157356,
      "loss": 0.2985,
      "step": 66000
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00014919396273145084,
      "loss": 0.2979,
      "step": 67000
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00014911917879669812,
      "loss": 0.2967,
      "step": 68000
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00014904109972251249,
      "loss": 0.2976,
      "step": 69000
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00014895972923118116,
      "loss": 0.2972,
      "step": 70000
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00014887507120190407,
      "loss": 0.2965,
      "step": 71000
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00014878712967060896,
      "loss": 0.2956,
      "step": 72000
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00014869590882975894,
      "loss": 0.2954,
      "step": 73000
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00014860141302815264,
      "loss": 0.2952,
      "step": 74000
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00014850364677071684,
      "loss": 0.2955,
      "step": 75000
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00014840261471829187,
      "loss": 0.2948,
      "step": 76000
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00014829832168740913,
      "loss": 0.2945,
      "step": 77000
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001481907726500617,
      "loss": 0.2938,
      "step": 78000
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00014807997273346728,
      "loss": 0.2936,
      "step": 79000
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00014796592721982364,
      "loss": 0.2932,
      "step": 80000
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001478486415460569,
      "loss": 0.2938,
      "step": 81000
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00014772812130356235,
      "loss": 0.2931,
      "step": 82000
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00014760437223793778,
      "loss": 0.2933,
      "step": 83000
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00014747740024870972,
      "loss": 0.2924,
      "step": 84000
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00014734721138905203,
      "loss": 0.2915,
      "step": 85000
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00014721381186549743,
      "loss": 0.2916,
      "step": 86000
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00014707720803764163,
      "loss": 0.2921,
      "step": 87000
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00014693740641784,
      "loss": 0.2914,
      "step": 88000
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00014679441367089737,
      "loss": 0.292,
      "step": 89000
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00014664823661374998,
      "loss": 0.2909,
      "step": 90000
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001464988822151407,
      "loss": 0.2913,
      "step": 91000
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00014634635759528682,
      "loss": 0.2905,
      "step": 92000
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00014619067002554046,
      "loss": 0.2911,
      "step": 93000
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.000146031826928042,
      "loss": 0.2901,
      "step": 94000
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00014586983587536634,
      "loss": 0.2902,
      "step": 95000
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001457047045901617,
      "loss": 0.2897,
      "step": 96000
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001455364409447816,
      "loss": 0.29,
      "step": 97000
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00014536505296090948,
      "loss": 0.2891,
      "step": 98000
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00014519054880917634,
      "loss": 0.2896,
      "step": 99000
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00014501293680877115,
      "loss": 0.2887,
      "step": 100000
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00014483222542704434,
      "loss": 0.2891,
      "step": 101000
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00014464842327910406,
      "loss": 0.2892,
      "step": 102000
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00014446153912740542,
      "loss": 0.2885,
      "step": 103000
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00014427158188133295,
      "loss": 0.2888,
      "step": 104000
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00014407856059677564,
      "loss": 0.288,
      "step": 105000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001438824844756953,
      "loss": 0.2877,
      "step": 106000
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00014368336286568802,
      "loss": 0.2875,
      "step": 107000
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00014348120525953817,
      "loss": 0.2879,
      "step": 108000
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001432760212947663,
      "loss": 0.287,
      "step": 109000
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00014306782075316932,
      "loss": 0.2875,
      "step": 110000
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00014285661356035442,
      "loss": 0.287,
      "step": 111000
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001426424097852657,
      "loss": 0.2868,
      "step": 112000
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00014242521963970427,
      "loss": 0.2868,
      "step": 113000
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00014220505347784133,
      "loss": 0.287,
      "step": 114000
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00014198192179572467,
      "loss": 0.2863,
      "step": 115000
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00014175583523077823,
      "loss": 0.2867,
      "step": 116000
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00014152680456129484,
      "loss": 0.2864,
      "step": 117000
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001412948407059227,
      "loss": 0.2861,
      "step": 118000
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00014105995472314447,
      "loss": 0.2865,
      "step": 119000
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00014082215781075044,
      "loss": 0.2857,
      "step": 120000
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00014058146130530442,
      "loss": 0.2857,
      "step": 121000
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00014033787668160343,
      "loss": 0.2852,
      "step": 122000
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001400914155521306,
      "loss": 0.2851,
      "step": 123000
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00013984208966650165,
      "loss": 0.2856,
      "step": 124000
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00013958991091090456,
      "loss": 0.2849,
      "step": 125000
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00013933489130753317,
      "loss": 0.2845,
      "step": 126000
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00013907704301401386,
      "loss": 0.2844,
      "step": 127000
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00013881637832282597,
      "loss": 0.2852,
      "step": 128000
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00013855290966071585,
      "loss": 0.2841,
      "step": 129000
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00013828664958810441,
      "loss": 0.2843,
      "step": 130000
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00013801761079848836,
      "loss": 0.2834,
      "step": 131000
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00013774580611783485,
      "loss": 0.2842,
      "step": 132000
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00013747124850397037,
      "loss": 0.2832,
      "step": 133000
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00013719395104596277,
      "loss": 0.2836,
      "step": 134000
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00013691392696349726,
      "loss": 0.2838,
      "step": 135000
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001366311896062463,
      "loss": 0.2835,
      "step": 136000
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00013634575245323311,
      "loss": 0.2835,
      "step": 137000
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00013605762911218905,
      "loss": 0.2831,
      "step": 138000
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00013576683331890498,
      "loss": 0.2829,
      "step": 139000
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001354733789365763,
      "loss": 0.2825,
      "step": 140000
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00013517727995514226,
      "loss": 0.2828,
      "step": 141000
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00013487855049061875,
      "loss": 0.2826,
      "step": 142000
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00013457720478442554,
      "loss": 0.2826,
      "step": 143000
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00013427325720270729,
      "loss": 0.2824,
      "step": 144000
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00013396672223564855,
      "loss": 0.2822,
      "step": 145000
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00013365761449678324,
      "loss": 0.2819,
      "step": 146000
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00013334594872229763,
      "loss": 0.282,
      "step": 147000
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00013303173977032815,
      "loss": 0.2815,
      "step": 148000
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00013271500262025278,
      "loss": 0.2815,
      "step": 149000
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00013239575237197705,
      "loss": 0.2818,
      "step": 150000
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00013207400424521432,
      "loss": 0.2818,
      "step": 151000
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001317497735787599,
      "loss": 0.2817,
      "step": 152000
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001314230758297601,
      "loss": 0.2817,
      "step": 153000
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.000131093926572975,
      "loss": 0.2809,
      "step": 154000
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00013076234150003634,
      "loss": 0.281,
      "step": 155000
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00013042833641869922,
      "loss": 0.2816,
      "step": 156000
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00013009192725208846,
      "loss": 0.2812,
      "step": 157000
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00012975313003793962,
      "loss": 0.2814,
      "step": 158000
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00012941196092783439,
      "loss": 0.2803,
      "step": 159000
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00012906843618643046,
      "loss": 0.2813,
      "step": 160000
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00012872257219068633,
      "loss": 0.2809,
      "step": 161000
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00012837438542908053,
      "loss": 0.2802,
      "step": 162000
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00012802389250082536,
      "loss": 0.2806,
      "step": 163000
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001276711101150758,
      "loss": 0.2806,
      "step": 164000
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001273160550901328,
      "loss": 0.2801,
      "step": 165000
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00012695874435264155,
      "loss": 0.2797,
      "step": 166000
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00012659919493678443,
      "loss": 0.2801,
      "step": 167000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001262374239834691,
      "loss": 0.28,
      "step": 168000
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00012587344873951116,
      "loss": 0.2797,
      "step": 169000
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001255072865568121,
      "loss": 0.2798,
      "step": 170000
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00012513895489153193,
      "loss": 0.2798,
      "step": 171000
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00012476847130325712,
      "loss": 0.2792,
      "step": 172000
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001243958534541634,
      "loss": 0.2795,
      "step": 173000
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00012402111910817367,
      "loss": 0.2796,
      "step": 174000
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00012364428613011134,
      "loss": 0.2785,
      "step": 175000
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001232653724848485,
      "loss": 0.2793,
      "step": 176000
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001228843962364495,
      "loss": 0.2787,
      "step": 177000
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001225013755473098,
      "loss": 0.2797,
      "step": 178000
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00012211632867729008,
      "loss": 0.2791,
      "step": 179000
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00012172927398284578,
      "loss": 0.2787,
      "step": 180000
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00012134022991615195,
      "loss": 0.279,
      "step": 181000
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00012094921502422359,
      "loss": 0.2788,
      "step": 182000
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00012055624794803143,
      "loss": 0.2787,
      "step": 183000
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00012016134742161328,
      "loss": 0.2789,
      "step": 184000
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001197645322711809,
      "loss": 0.2783,
      "step": 185000
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00011936582141422247,
      "loss": 0.2782,
      "step": 186000
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00011896523385860084,
      "loss": 0.2783,
      "step": 187000
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00011856278870164717,
      "loss": 0.2775,
      "step": 188000
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00011815850512925069,
      "loss": 0.2779,
      "step": 189000
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00011775240241494394,
      "loss": 0.2777,
      "step": 190000
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00011734449991898394,
      "loss": 0.2772,
      "step": 191000
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00011693481708742928,
      "loss": 0.2778,
      "step": 192000
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000116523373451213,
      "loss": 0.2773,
      "step": 193000
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001161101886252115,
      "loss": 0.2779,
      "step": 194000
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00011569528230730951,
      "loss": 0.2778,
      "step": 195000
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00011527867427746085,
      "loss": 0.2777,
      "step": 196000
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00011486038439674568,
      "loss": 0.2772,
      "step": 197000
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00011444043260642344,
      "loss": 0.2773,
      "step": 198000
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00011401883892698241,
      "loss": 0.2775,
      "step": 199000
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00011359562345718505,
      "loss": 0.2766,
      "step": 200000
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00011317080637310989,
      "loss": 0.2763,
      "step": 201000
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00011274440792718983,
      "loss": 0.2765,
      "step": 202000
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001123164484472464,
      "loss": 0.2768,
      "step": 203000
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00011188694833552081,
      "loss": 0.276,
      "step": 204000
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00011145592806770127,
      "loss": 0.2765,
      "step": 205000
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00011102340819194689,
      "loss": 0.2763,
      "step": 206000
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00011058940932790795,
      "loss": 0.2763,
      "step": 207000
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00011015395216574308,
      "loss": 0.2764,
      "step": 208000
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00010971705746513272,
      "loss": 0.276,
      "step": 209000
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00010927874605428955,
      "loss": 0.2759,
      "step": 210000
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00010883903882896555,
      "loss": 0.276,
      "step": 211000
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00010839795675145566,
      "loss": 0.2756,
      "step": 212000
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00010795552084959867,
      "loss": 0.2756,
      "step": 213000
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00010751175221577465,
      "loss": 0.2753,
      "step": 214000
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00010706667200589931,
      "loss": 0.2762,
      "step": 215000
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00010662030143841562,
      "loss": 0.2754,
      "step": 216000
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00010617266179328208,
      "loss": 0.2755,
      "step": 217000
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00010572377441095834,
      "loss": 0.2759,
      "step": 218000
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00010527366069138784,
      "loss": 0.2754,
      "step": 219000
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00010482234209297747,
      "loss": 0.2756,
      "step": 220000
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001043698401315748,
      "loss": 0.2752,
      "step": 221000
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00010391617637944215,
      "loss": 0.2751,
      "step": 222000
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00010346137246422819,
      "loss": 0.2745,
      "step": 223000
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00010300545006793699,
      "loss": 0.2747,
      "step": 224000
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00010254843092589434,
      "loss": 0.2746,
      "step": 225000
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00010209033682571144,
      "loss": 0.2749,
      "step": 226000
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00010163118960624632,
      "loss": 0.2739,
      "step": 227000
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001011710111565627,
      "loss": 0.2746,
      "step": 228000
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00010070982341488645,
      "loss": 0.2744,
      "step": 229000
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00010024764836755973,
      "loss": 0.2742,
      "step": 230000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.978450804799277e-05,
      "loss": 0.2737,
      "step": 231000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.932042453561354e-05,
      "loss": 0.2747,
      "step": 232000
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.885541995481517e-05,
      "loss": 0.2741,
      "step": 233000
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.838951647390106e-05,
      "loss": 0.2747,
      "step": 234000
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.792273630402829e-05,
      "loss": 0.2739,
      "step": 235000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.745510169814845e-05,
      "loss": 0.2748,
      "step": 236000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.698663494994703e-05,
      "loss": 0.2736,
      "step": 237000
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.65173583927804e-05,
      "loss": 0.2732,
      "step": 238000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.604729439861125e-05,
      "loss": 0.274,
      "step": 239000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.5576465376942e-05,
      "loss": 0.2739,
      "step": 240000
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.510489377374642e-05,
      "loss": 0.2723,
      "step": 241000
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.463260207039961e-05,
      "loss": 0.2732,
      "step": 242000
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.415961278260625e-05,
      "loss": 0.2742,
      "step": 243000
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.36859484593271e-05,
      "loss": 0.2729,
      "step": 244000
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.321163168170416e-05,
      "loss": 0.2732,
      "step": 245000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.273668506198407e-05,
      "loss": 0.2728,
      "step": 246000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.226113124244006e-05,
      "loss": 0.2726,
      "step": 247000
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.178499289429266e-05,
      "loss": 0.273,
      "step": 248000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.130829271662884e-05,
      "loss": 0.2731,
      "step": 249000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.083105343531967e-05,
      "loss": 0.2726,
      "step": 250000
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.035329780193725e-05,
      "loss": 0.2728,
      "step": 251000
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.987504859266978e-05,
      "loss": 0.2727,
      "step": 252000
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.939632860723587e-05,
      "loss": 0.272,
      "step": 253000
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.891716066779766e-05,
      "loss": 0.2726,
      "step": 254000
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.843756761787259e-05,
      "loss": 0.272,
      "step": 255000
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.79575723212447e-05,
      "loss": 0.2716,
      "step": 256000
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.747719766087431e-05,
      "loss": 0.272,
      "step": 257000
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.69964665378073e-05,
      "loss": 0.2723,
      "step": 258000
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.651540187008339e-05,
      "loss": 0.2719,
      "step": 259000
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.603402659164335e-05,
      "loss": 0.2723,
      "step": 260000
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.555236365123586e-05,
      "loss": 0.2714,
      "step": 261000
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.507043601132336e-05,
      "loss": 0.2719,
      "step": 262000
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.458826664698743e-05,
      "loss": 0.2717,
      "step": 263000
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.410587854483341e-05,
      "loss": 0.2719,
      "step": 264000
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.362329470189466e-05,
      "loss": 0.2715,
      "step": 265000
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.314053812453604e-05,
      "loss": 0.272,
      "step": 266000
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.265763182735732e-05,
      "loss": 0.271,
      "step": 267000
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.217459883209592e-05,
      "loss": 0.2715,
      "step": 268000
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.169146216652924e-05,
      "loss": 0.2713,
      "step": 269000
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.12082448633772e-05,
      "loss": 0.2708,
      "step": 270000
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.072496995920378e-05,
      "loss": 0.2708,
      "step": 271000
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.024166049331909e-05,
      "loss": 0.2706,
      "step": 272000
    },
    {
      "epoch": 1.62,
      "learning_rate": 7.975833950668091e-05,
      "loss": 0.2705,
      "step": 273000
    },
    {
      "epoch": 1.62,
      "learning_rate": 7.927503004079623e-05,
      "loss": 0.2707,
      "step": 274000
    },
    {
      "epoch": 1.63,
      "learning_rate": 7.87917551366228e-05,
      "loss": 0.2705,
      "step": 275000
    },
    {
      "epoch": 1.64,
      "learning_rate": 7.830853783347074e-05,
      "loss": 0.2703,
      "step": 276000
    },
    {
      "epoch": 1.64,
      "learning_rate": 7.782540116790411e-05,
      "loss": 0.2703,
      "step": 277000
    },
    {
      "epoch": 1.65,
      "learning_rate": 7.734236817264267e-05,
      "loss": 0.2697,
      "step": 278000
    },
    {
      "epoch": 1.65,
      "learning_rate": 7.685946187546397e-05,
      "loss": 0.2699,
      "step": 279000
    },
    {
      "epoch": 1.66,
      "learning_rate": 7.637670529810532e-05,
      "loss": 0.2703,
      "step": 280000
    },
    {
      "epoch": 1.67,
      "learning_rate": 7.589412145516656e-05,
      "loss": 0.2701,
      "step": 281000
    },
    {
      "epoch": 1.67,
      "learning_rate": 7.541173335301255e-05,
      "loss": 0.2704,
      "step": 282000
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.492956398867663e-05,
      "loss": 0.2701,
      "step": 283000
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.444763634876413e-05,
      "loss": 0.2703,
      "step": 284000
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.396597340835663e-05,
      "loss": 0.2701,
      "step": 285000
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.34845981299166e-05,
      "loss": 0.2703,
      "step": 286000
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.300353346219268e-05,
      "loss": 0.2694,
      "step": 287000
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.252280233912569e-05,
      "loss": 0.2695,
      "step": 288000
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.20424276787553e-05,
      "loss": 0.2694,
      "step": 289000
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.156243238212739e-05,
      "loss": 0.2701,
      "step": 290000
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.108283933220234e-05,
      "loss": 0.2696,
      "step": 291000
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.060367139276414e-05,
      "loss": 0.2696,
      "step": 292000
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.012495140733023e-05,
      "loss": 0.2694,
      "step": 293000
    },
    {
      "epoch": 1.74,
      "learning_rate": 6.964670219806273e-05,
      "loss": 0.2691,
      "step": 294000
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.91689465646803e-05,
      "loss": 0.2686,
      "step": 295000
    },
    {
      "epoch": 1.75,
      "learning_rate": 6.869170728337114e-05,
      "loss": 0.2692,
      "step": 296000
    },
    {
      "epoch": 1.76,
      "learning_rate": 6.82150071057073e-05,
      "loss": 0.2691,
      "step": 297000
    },
    {
      "epoch": 1.77,
      "learning_rate": 6.773886875755992e-05,
      "loss": 0.2684,
      "step": 298000
    },
    {
      "epoch": 1.77,
      "learning_rate": 6.726331493801593e-05,
      "loss": 0.2688,
      "step": 299000
    },
    {
      "epoch": 1.78,
      "learning_rate": 6.678836831829584e-05,
      "loss": 0.2693,
      "step": 300000
    }
  ],
  "max_steps": 500000,
  "num_train_epochs": 3,
  "total_flos": 1.1529473128203587e+21,
  "trial_name": null,
  "trial_params": null
}